Paraphase question

[goldstache]

Had a question about deriving the signal feed to the second triode of a paraphase inverter.


I was adding a power tube bias trem to a gibson ga40 I built up single channel (non trem channel).


In my travels with that amp, it’s seen all sorts of setups. And the stock PI of ga40 uses the divider network to second PI triode as the power tube grid leak as well.
So if I wanted power tube bias trem would have to whack something else on there. At least at a glance.


At some point I used a 5e9 trem setup to wiggle PI.


So the question is regarding the ac signal to the second trios of PI. If I use the cross plate divider that ampeg and others have employed, seems like there’s more idle noise in that stage. Whereas you tap that ac signal post coupler with vdivider to gnd. Noise is less.


Somebody enlighten me here?
Thanks

[SEL49]

The J12B uses a slightly different paraphase inverter and it also has trem signal applied to the grids of the output tubes. I would modify the ga-40 pi to look like the j12b pi. It's a fairly simple mod on paper but could be a bit of a challenge in the chassis. Take a look and see if its worth trying...

     https://sluckeyamps.com/RCA/Ampeg_J12B.pdf

[goldstache]

Awesome thanks!
Honestly that’s what pushed me here. I cobbled in an ampeg r12-rb paraphase in, and found it to be noisy.
(Paraphase with input to 2nd triode derived from plate to plate divider, and slightly different cathode network.)


And that’s where the chassis is presently.


Noticed that it was noisier with the plate divided setup over the post coupling cap signal tap.


I get that they are different circuits, but AC wasn’t much different headed to power glass.


We see shunt caps across LTPI often. And even the ampeg r12-rb has a little shunt cap on the plate of first PI triode.


So I’m kinda thinking that the post PI coupling cap divider may be a more stable design.


The cross plate divider seemed to really up the idle noise in that stage.
So got me wondering why, if at all.


I realized that ground scheme plays a role here.
But rejection different between those two topologies?